Instructions to use Danielbrdz/BarcenasNahuatl-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Danielbrdz/BarcenasNahuatl-4b with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Danielbrdz/BarcenasNahuatl-4b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Danielbrdz/BarcenasNahuatl-4b")
model = AutoModelForCausalLM.from_pretrained("Danielbrdz/BarcenasNahuatl-4b", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Danielbrdz/BarcenasNahuatl-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Danielbrdz/BarcenasNahuatl-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Danielbrdz/BarcenasNahuatl-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Danielbrdz/BarcenasNahuatl-4b

SGLang

How to use Danielbrdz/BarcenasNahuatl-4b with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Danielbrdz/BarcenasNahuatl-4b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Danielbrdz/BarcenasNahuatl-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Danielbrdz/BarcenasNahuatl-4b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Danielbrdz/BarcenasNahuatl-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Danielbrdz/BarcenasNahuatl-4b with Docker Model Runner:
```
docker model run hf.co/Danielbrdz/BarcenasNahuatl-4b
```

Barcenas Náhuatl 4B

Basado en Qwen 3 4B Instruct 2507 y entrenado con el dataset Barcenas Náhuatl.

El objetivo de este LLM es tener un modelo pequeño que se pueda ejecutar en la mayoría de equipos y tenga la capacidad de más o menos escribir náhuatl.

Todo esto es posible gracias al dataset Barcenas Náhuatl, hecho por un servidor, que contiene 100,000 ejemplos de español a náhuatl clásico.

Usando el pequeño benchmark que realicé para evaluar el náhuatl de un LLM, el modelo sin fine-tuning sacó un 14.9, mientras que esta versión refinada sacó un 33.9, prácticamente igualando a Kimi K2 Instruct 0905, pero con un tamaño mucho menor.

Barcenas Nahuatl 4B

Based on Qwen 3 4B Instruct 2507 and trained with the Barcenas Náhuatl dataset.

The goal of this LLM is to have a small model that can run on most computers and has the ability to more or less write Nahuatl.

All of this is possible thanks to the server-side Barcenas Nahuatl dataset, which contains 100,000 examples of Spanish to Classical Nahuatl.

Using the small benchmark I did to evaluate the language of an LLM, the model without fine-tuning scored a 14.9, while this refined version scored a 33.9, practically equaling the Kimi K2 Instruct 0905, but with a much smaller size.

Made with ❤️ in Guadalupe, Nuevo Leon, Mexico 🇲🇽

Downloads last month: 6

Safetensors

Model size

4B params

Tensor type

F16

Model tree for Danielbrdz/BarcenasNahuatl-4b

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(2085)

this model

Danielbrdz
/

BarcenasNahuatl-4b

Model tree for Danielbrdz/BarcenasNahuatl-4b

Dataset used to train Danielbrdz/BarcenasNahuatl-4b